Codebook

codebook_data <- import("positive_results_in_registered_reports_masterfile_data.xlsx")

# Read codebook from second sheet in Excel file.
dict <- import("positive_results_in_registered_reports_masterfile_data.xlsx", sheet = "data_dictionary")

# Add variable labels to data (from Excel file)
var_label(codebook_data) <- dict_to_list(dict)

# Add value labels to data (manually)
val_labels(codebook_data$include_in_analysis) <- c("excluded" = 0, "included" = 1)
val_labels(codebook_data$is_RR) <- c("SR" = 0, "RR"= 1)
val_labels(codebook_data$support_binary) <- c("no support" = 0, "partial or full support" = 1)
val_labels(codebook_data$is_replication) <- c("is not a replication" = 0, "is a replication" = 1)
val_labels(codebook_data$contains_replic) <- c("does not contain 'replic'" = 0, "contains 'replic'" = 1)
val_labels(codebook_data$hyp_from_fulltext) <- c("hypothesis from abstract alone" = 0, "hypothesis from full text" = 1)
val_labels(codebook_data$result_from_fulltext) <- c("result from abstract alone" = 0, "result from full text" = 1)
val_labels(codebook_data$coding_round) <- c("RR pilot" = 0,
                                            "main coding round" = 1,
                                            "SR replacements and late RR additions" = 2,
                                            "second SR replacement" = 3,
                                            "RR protocol replacement" = 4)
val_labels(codebook_data$uses_AS_coding) <- c("uses MS coding" = 0, "uses AS coding" = 1)
val_labels(codebook_data$coded_by_MS) <- c("not coded by MS" = 0, "coded by MS" = 1)
val_labels(codebook_data$evaluation_certainty_MS) <- c("not certain" = 1,
                                                       2,
                                                       3,
                                                       4,
                                                       "very certain" = 5)
val_labels(codebook_data$hypothesis_from_fulltext_MS) <- c("hypothesis from abstract alone" = 0, "hypothesis from full text" = 1)
val_labels(codebook_data$result_from_fulltext_MS) <- c("result from abstract alone" = 0, "result from full text" = 1)
val_labels(codebook_data$fulltext_available_MS) <- c("full text not available" = 0, "full text available" = 1)
val_labels(codebook_data$is_replication_old_MS) <- c("does not contain a replication" = 0, "contains a replication" = 1)
val_labels(codebook_data$is_original_old_MS) <- c("does not contain original work" = 0, "contains original work" = 1)
val_labels(codebook_data$coded_by_AS) <- c("not coded by AS" = 0, "coded by AS" = 1)
val_labels(codebook_data$evaluation_certainty_AS) <- c("not certain" = 1,
                                                       2,
                                                       3,
                                                       4,
                                                       "very certain" = 5)
val_labels(codebook_data$hypothesis_from_fulltext_AS) <- c("hypothesis from abstract alone" = 0, "hypothesis from full text" = 1)
val_labels(codebook_data$result_from_fulltext_AS) <- c("result from abstract alone" = 0, "result from full text" = 1)
val_labels(codebook_data$fulltext_available_AS) <- c("full text not available" = 0, "full text available" = 1)
val_labels(codebook_data$is_replication_old_AS) <- c("does not contain a replication" = 0, "contains a replication" = 1)
val_labels(codebook_data$is_original_old_AS) <- c("does not contain original work" = 0, "contains original work" = 1)
val_labels(codebook_data$is_replication_new_coded_by_AS) <- c("replication status not coded by AS" = 0, "replication status coded by AS" = 1)
val_labels(codebook_data$is_replication_new_AS) <- c("is not a replication" = 0, "is a replication" = 1)
val_labels(codebook_data$is_replication_new_coded_by_DL) <- c("replication status not coded by DL" = 0, "replication status coded by DL" = 1)
val_labels(codebook_data$is_replication_new_DL) <- c("is not a replication" = 0, "is a replication" = 1)
val_labels(codebook_data$is_replication_special_issue) <- c("not part of replication special issue" = 0, "part of replication special issue" = 1)
val_labels(codebook_data$hypothesis_intro_contains_replic) <- c("hypothesis intro does not contain 'replic'" = 0, "hypothesis intro contains 'replic'" = 1)


# Add meta data
metadata(codebook_data)$name <- "Positive Results in Standard vs Registered Reports"
metadata(codebook_data)$description <- "Coded variables to compare the proportion of papers that find support for their first tested hypothesis between Registered Reports and standard reports in Psychology"
metadata(codebook_data)$creator <- list(
      "@type" = "Person",
      givenName = "Anne", familyName = "Scheel",
      affiliation = list("@type" = "Organization",
        name = "Eindhoven University of Technology"),
      "@type" = "Person",
      givenName = " Mitchell", familiyName = "Schijen",
      "@type" = "Person",
      givenName = "Daniël", familiyName = "Lakens")
metadata(codebook_data)$temporalCoverage <- "Winter 2018"

#Export codebook
rio::export(codebook_data, "positive_results_in_registered_reports_data.rds") # to R data structure file

codebook(codebook_data)

knitr::asis_output(data_info)

Metadata

Description

if (exists("name", meta)) {
  glue::glue(
    "__Dataset name__: {name}",
    .envir = meta)
}

Dataset name: Positive Results in Standard vs Registered Reports

cat(description)

Coded variables to compare the proportion of papers that find support for their first tested hypothesis between Registered Reports and standard reports in Psychology

Metadata for search engines

• Temporal Coverage: Winter 2018

• Date published: 2020-01-13

• Creator:

  • @type: Person
  • givenName: Anne
  • familyName: Scheel

  • affiliation:

    • @type: Organization
    • name: Eindhoven University of Technology
  • @type: Person
  • givenName: Mitchell
  • familiyName: Schijen
  • @type: Person
  • givenName: Daniël

  • familiyName: Lakens

meta <- meta[setdiff(names(meta),
                     c("creator", "datePublished", "identifier",
                       "url", "citation", "spatialCoverage", 
                       "temporalCoverage", "description", "name"))]
pander::pander(meta)

• keywords: id, include_in_analysis, exclusion_crit, is_RR, support, support_binary, is_replication, contains_replic, RR_hyp_intro_abstract, RR_hyp_intro_fulltext, hypothesis_quote, result_quote, hyp_from_fulltext, result_from_fulltext, doi, title, authors, year, journal, abstract, keywords, coding_round, combining_coding_notes, uses_AS_coding, coded_by_MS, support_orig_MS, support_revised_MS, evaluation_certainty_MS, hypothesis_MS, hypothesis_introduction_MS, finding_MS, conclusion_MS, hypothesis_from_fulltext_MS, result_from_fulltext_MS, fulltext_available_MS, is_replication_old_MS, is_original_old_MS, time_spent_MS, individual_coding_notes_MS, coded_by_AS, support_orig_AS, support_revised_AS, evaluation_certainty_AS, hypothesis_AS, hypothesis_introduction_AS, finding_AS, conclusion_AS, hypothesis_from_fulltext_AS, result_from_fulltext_AS, fulltext_available_AS, is_replication_old_AS, is_original_old_AS, time_spent_AS, individual_coding_notes_AS, is_replication_new_combining_notes, is_replication_new_coded_by_AS, is_replication_new_AS, replic_notes_AS, is_replication_new_coded_by_DL, is_replication_new_DL, replic_notes_DL, is_replication_special_issue and hypothesis_intro_contains_replic

knitr::asis_output(survey_overview)

Variables

if (detailed_variables || detailed_scales) {
  knitr::asis_output(paste0(scales_items, sep = "\n\n\n", collapse = "\n\n\n"))
}

id

ID number

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	min	median	max	mean	sd	hist
id	ID number	numeric	0	1	1	122	243	122	70.29225	▇▇▇▇▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

include_in_analysis

whether paper is included in final analysis

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
include_in_analysis	whether paper is included in final analysis	haven_labelled	0. excluded, 1. included	0	1	0	1	1	0.9176955	0.2753954	2	▁▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• excluded: 0
• included: 1

exclusion_crit

reason for exclusion

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

223 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
exclusion_crit	reason for exclusion	character	223	0.0823045	5	0	3	24	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

is_RR

Registered Report (RR) or standard report (SR)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_RR	Registered Report (RR) or standard report (SR)	haven_labelled	0. SR, 1. RR	0	1	0	0	1	0.345679	0.4765708	2	▇▁▁▁▁▁▁▅

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• SR: 0
• RR: 1

support

if the coded hypothesis received no support, partial support, or full support

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
support	if the coded hypothesis received no support, partial support, or full support	character	0	1	4	0	7	15	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

support_binary

whether the coded hypothesis was at least partially supported or not

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

20 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
support_binary	whether the coded hypothesis was at least partially supported or not	haven_labelled	0. no support, 1. partial or full support	20	0.9176955	0	1	1	0.793722	0.4055427	2	▂▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• no support: 0
• partial or full support: 1

is_replication

whether the coded hypothesis is a direct replication of previous work

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

2 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication	whether the coded hypothesis is a direct replication of previous work	haven_labelled	0. is not a replication, 1. is a replication	2	0.9917695	0	0	1	0.2323651	0.4232196	2	▇▁▁▁▁▁▁▂

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• is not a replication: 0
• is a replication: 1

contains_replic

whether the full text of the paper (excl. references) contains the string ‘replic’

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
contains_replic	whether the full text of the paper (excl. references) contains the string ‘replic’	haven_labelled	0. does not contain ‘replic’, 1. contains ‘replic’	0	1	0	1	1	0.6419753	0.480409	2	▅▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• does not contain ‘replic’: 0
• contains ‘replic’: 1

RR_hyp_intro_abstract

phrase used to introduce/mark the coded hypothesis in RRs; from abstract

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

194 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
RR_hyp_intro_abstract	phrase used to introduce/mark the coded hypothesis in RRs; from abstract	character	194	0.2016461	38	0	13	43	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

RR_hyp_intro_fulltext

phrase used to introduce/mark the coded hypothesis in RRs; from full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

218 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
RR_hyp_intro_fulltext	phrase used to introduce/mark the coded hypothesis in RRs; from full text	character	218	0.1028807	23	0	2	54	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

hypothesis_quote

quote of the coded hypothesis

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

20 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
hypothesis_quote	quote of the coded hypothesis	character	20	0.9176955	223	0	83	1464	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

result_quote

quote of the coded finding or conclusion

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

20 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
result_quote	quote of the coded finding or conclusion	character	20	0.9176955	223	0	27	1872	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

hyp_from_fulltext

whether hypothesis was coded from abstract or full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

18 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
hyp_from_fulltext	whether hypothesis was coded from abstract or full text	haven_labelled	0. hypothesis from abstract alone, 1. hypothesis from full text	18	0.9259259	0	0	1	0.2844444	0.4521553	2	▇▁▁▁▁▁▁▃

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• hypothesis from abstract alone: 0
• hypothesis from full text: 1

result_from_fulltext

whether finding/conclusion was coded from abstract or full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

19 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
result_from_fulltext	whether finding/conclusion was coded from abstract or full text	haven_labelled	0. result from abstract alone, 1. result from full text	19	0.9218107	0	0	1	0.34375	0.4760226	2	▇▁▁▁▁▁▁▅

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• result from abstract alone: 0
• result from full text: 1

doi

digital object identifier (DOI)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

1 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
doi	digital object identifier (DOI)	character	1	0.9958848	242	0	14	38	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

title

title

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
title	title	character	0	1	243	0	34	230	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

authors

author(s) of the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
authors	author(s) of the paper	character	0	1	243	0	12	1321	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

year

publication year

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	min	median	max	mean	sd	hist
year	publication year	numeric	0	1	2013	2016	2018	2015.844	1.63896	▇▃▅▆▆

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

journal

journal the paper is published in

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

1 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
journal	journal the paper is published in	character	1	0.9958848	117	0	6	66	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

abstract

abstract

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

20 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
abstract	abstract	character	20	0.9176955	223	0	604	3107	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

keywords

author keywords from meta-data

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

43 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
keywords	author keywords from meta-data	character	43	0.8230453	200	0	26	190	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

coding_round

when the paper was coded (during piloting, in the main round, or as a replacement/late addition)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
coding_round	when the paper was coded (during piloting, in the main round, or as a replacement/late addition)	haven_labelled	0. RR pilot, 1. main coding round, 2. SR replacements and late RR additions, 3. second SR replacement, 4. RR protocol replacement	0	1	0	1	4	1.024691	0.3740103	5	▁▇▁▁▁▁▁▁

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• RR pilot: 0
• main coding round: 1
• SR replacements and late RR additions: 2
• second SR replacement: 3
• RR protocol replacement: 4

combining_coding_notes

Comments on how MS’ and AS’ codings were combined if there was disagreement

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

170 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
combining_coding_notes	Comments on how MS’ and AS’ codings were combined if there was disagreement	character	170	0.3004115	40	0	47	498	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

uses_AS_coding

whether AS’ or MS’ coding was used for final analysis

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

3 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
uses_AS_coding	whether AS’ or MS’ coding was used for final analysis	haven_labelled	0. uses MS coding, 1. uses AS coding	3	0.9876543	0	0	1	0.1833333	0.3877482	2	▇▁▁▁▁▁▁▂

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• uses MS coding: 0
• uses AS coding: 1

coded_by_MS

whether MS coded the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
coded_by_MS	whether MS coded the paper	haven_labelled	0. not coded by MS, 1. coded by MS	0	1	0	1	1	0.9876543	0.110651	2	▁▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• not coded by MS: 0
• coded by MS: 1

support_orig_MS

MS: whether hypothesis was supported (original verdict)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
support_orig_MS	MS: whether hypothesis was supported (original verdict)	character	0	1	4	0	7	15	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

support_revised_MS

MS: whether hypothesis was supported (revised verdict after discussion with MS)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

21 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
support_revised_MS	MS: whether hypothesis was supported (revised verdict after discussion with MS)	character	21	0.9135802	4	0	7	15	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

evaluation_certainty_MS

MS’ coding certainty

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

13 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
evaluation_certainty_MS	MS’ coding certainty	haven_labelled	1. not certain, 2. , 3. , 4. , 5. very certain	13	0.9465021	1	4	5	3.582609	1.344454	5	▃▂▁▃▁▇▁▆

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• not certain: 1
• 2
• 3
• 4
• very certain: 5

hypothesis_MS

MS: hypothesis quote

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

16 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
hypothesis_MS	MS: hypothesis quote	character	16	0.9341564	227	0	83	1464	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

hypothesis_introduction_MS

MS: hypothesis introduction phrase

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

32 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
hypothesis_introduction_MS	MS: hypothesis introduction phrase	character	32	0.8683128	94	0	7	126	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

finding_MS

MS: finding quote

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

87 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
finding_MS	MS: finding quote	character	87	0.6419753	156	0	27	1402	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

conclusion_MS

MS: conclusion quote

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

164 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
conclusion_MS	MS: conclusion quote	character	164	0.3251029	79	0	27	1729	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

hypothesis_from_fulltext_MS

MS: whether hypothesis was coded from abstract or full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

18 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
hypothesis_from_fulltext_MS	MS: whether hypothesis was coded from abstract or full text	haven_labelled	0. hypothesis from abstract alone, 1. hypothesis from full text	18	0.9259259	0	0	1	0.3333333	0.4724556	2	▇▁▁▁▁▁▁▃

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• hypothesis from abstract alone: 0
• hypothesis from full text: 1

result_from_fulltext_MS

MS: whether finding/conclusion was coded from abstract or full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

19 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
result_from_fulltext_MS	MS: whether finding/conclusion was coded from abstract or full text	haven_labelled	0. result from abstract alone, 1. result from full text	19	0.9218107	0	0	1	0.3794643	0.4863405	2	▇▁▁▁▁▁▁▅

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• result from abstract alone: 0
• result from full text: 1

fulltext_available_MS

MS: whether full text was available during coding

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

2 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
fulltext_available_MS	MS: whether full text was available during coding	haven_labelled	0. full text not available, 1. full text available	2	0.9917695	0	1	1	0.8755187	0.3308169	2	▁▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• full text not available: 0
• full text available: 1

is_replication_old_MS

MS: whether paper contains a replication (old variable, not used!)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

6 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_old_MS	MS: whether paper contains a replication (old variable, not used!)	haven_labelled	0. does not contain a replication, 1. contains a replication	6	0.9753086	0	0	1	0.3417722	0.4753075	2	▇▁▁▁▁▁▁▅

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• does not contain a replication: 0
• contains a replication: 1

is_original_old_MS

MS: whether paper contains original research (old variable, not used!)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

6 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_original_old_MS	MS: whether paper contains original research (old variable, not used!)	haven_labelled	0. does not contain original work, 1. contains original work	6	0.9753086	0	1	1	0.8396624	0.3676954	2	▂▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• does not contain original work: 0
• contains original work: 1

time_spent_MS

MS: time spent on coding the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

## 27  unique, categorical values, so not shown.

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	min	median	max
time_spent_MS	MS: time spent on coding the paper	POSIXct	0	1	27	1899-12-31 00:02:00	1899-12-31 00:05:00	1899-12-31 00:30:00

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

individual_coding_notes_MS

MS: coding notes/comments

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

88 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
individual_coding_notes_MS	MS: coding notes/comments	character	88	0.6378601	153	0	26	842	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

coded_by_AS

whether AS coded the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
coded_by_AS	whether AS coded the paper	haven_labelled	0. not coded by AS, 1. coded by AS	0	1	0	0	1	0.3621399	0.481611	2	▇▁▁▁▁▁▁▅

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• not coded by AS: 0
• coded by AS: 1

support_orig_AS

AS: whether hypothesis was supported (original verdict)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

153 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
support_orig_AS	AS: whether hypothesis was supported (original verdict)	character	153	0.3703704	8	0	7	28	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

support_revised_AS

AS: whether hypothesis was supported (revised verdict after discussion with MS)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

161 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
support_revised_AS	AS: whether hypothesis was supported (revised verdict after discussion with MS)	character	161	0.3374486	5	0	7	15	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

evaluation_certainty_AS

AS’ coding certainty

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

232 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
evaluation_certainty_AS	AS’ coding certainty	haven_labelled	1. not certain, 2. , 3. , 4. , 5. very certain	232	0.0452675	2	4	5	4	0.8944272	5	▁▁▁▁▁▇▁▃

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• not certain: 1
• 2
• 3
• 4
• very certain: 5

hypothesis_AS

AS: hypothesis quote

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

164 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
hypothesis_AS	AS: hypothesis quote	character	164	0.3251029	79	0	26	593	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

hypothesis_introduction_AS

AS: hypothesis introduction phrase

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

169 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
hypothesis_introduction_AS	AS: hypothesis introduction phrase	character	169	0.3045267	40	0	3	54	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

finding_AS

AS: finding quote

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

199 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
finding_AS	AS: finding quote	character	199	0.18107	44	0	98	677	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

conclusion_AS

AS: conclusion quote

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

201 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
conclusion_AS	AS: conclusion quote	character	201	0.1728395	42	0	51	542	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

hypothesis_from_fulltext_AS

AS: whether hypothesis was coded from abstract or full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

168 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
hypothesis_from_fulltext_AS	AS: whether hypothesis was coded from abstract or full text	haven_labelled	0. hypothesis from abstract alone, 1. hypothesis from full text	168	0.308642	0	0	1	0.1466667	0.3561556	2	▇▁▁▁▁▁▁▂

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• hypothesis from abstract alone: 0
• hypothesis from full text: 1

result_from_fulltext_AS

AS: whether finding/conclusion was coded from abstract or full text

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

171 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
result_from_fulltext_AS	AS: whether finding/conclusion was coded from abstract or full text	haven_labelled	0. result from abstract alone, 1. result from full text	171	0.2962963	0	0	1	0.25	0.4360514	2	▇▁▁▁▁▁▁▂

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• result from abstract alone: 0
• result from full text: 1

fulltext_available_AS

AS: whether full text was available during coding

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

164 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
fulltext_available_AS	AS: whether full text was available during coding	haven_labelled	0. full text not available, 1. full text available	164	0.3251029	0	1	1	0.9367089	0.2450417	2	▁▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• full text not available: 0
• full text available: 1

is_replication_old_AS

AS: whether paper contains a replication (old variable, not used!)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

146 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_old_AS	AS: whether paper contains a replication (old variable, not used!)	haven_labelled	0. does not contain a replication, 1. contains a replication	146	0.399177	0	0	1	0.2680412	0.4452402	2	▇▁▁▁▁▁▁▃

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• does not contain a replication: 0
• contains a replication: 1

is_original_old_AS

AS: whether paper contains original research (old variable, not used!)

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

146 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_original_old_AS	AS: whether paper contains original research (old variable, not used!)	haven_labelled	0. does not contain original work, 1. contains original work	146	0.399177	0	1	1	0.8659794	0.3424442	2	▁▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• does not contain original work: 0
• contains original work: 1

time_spent_AS

AS: time spent on coding the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

## 29  unique, categorical values, so not shown.

knitr::opts_chunk$set(fig.height = old_height)

161 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	min	median	max
time_spent_AS	AS: time spent on coding the paper	POSIXct	161	0.3374486	29	1899-12-31 00:01:30	1899-12-31 00:04:00	1899-12-31 00:15:00

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

individual_coding_notes_AS

AS: coding notes/comments

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

202 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
individual_coding_notes_AS	AS: coding notes/comments	character	202	0.1687243	38	0	17	489	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

is_replication_new_combining_notes

Comments on how AS’ and DL’s new replication codings were combined if there was disagreement

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

235 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
is_replication_new_combining_notes	Comments on how AS’ and DL’s new replication codings were combined if there was disagreement	character	235	0.0329218	6	0	39	119	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

is_replication_new_coded_by_AS

whether AS coded replication status (new variable) for the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_new_coded_by_AS	whether AS coded replication status (new variable) for the paper	haven_labelled	0. replication status not coded by AS, 1. replication status coded by AS	0	1	1	1	1	1	0	2	▁▁▁▇▁▁▁▁

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• replication status not coded by AS: 0
• replication status coded by AS: 1

is_replication_new_AS

AS: whether coded hypothesis is a direct replication of previous work

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

1 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_new_AS	AS: whether coded hypothesis is a direct replication of previous work	haven_labelled	0. is not a replication, 1. is a replication	1	0.9958848	0	0	1	0.2272727	0.4199387	2	▇▁▁▁▁▁▁▂

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• is not a replication: 0
• is a replication: 1

replic_notes_AS

AS: coding notes/comments re new replication status variable

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

227 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
replic_notes_AS	AS: coding notes/comments re new replication status variable	character	227	0.0658436	16	0	31	575	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

is_replication_new_coded_by_DL

whether DL coded replication status (new variable) for the paper

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_new_coded_by_DL	whether DL coded replication status (new variable) for the paper	haven_labelled	0. replication status not coded by DL, 1. replication status coded by DL	0	1	0	1	1	0.5967078	0.491571	2	▆▁▁▁▁▁▁▇

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• replication status not coded by DL: 0
• replication status coded by DL: 1

is_replication_new_DL

DL: whether coded hypothesis is a direct replication of previous work

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

100 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_new_DL	DL: whether coded hypothesis is a direct replication of previous work	haven_labelled	0. is not a replication, 1. is a replication	100	0.5884774	0	0	1	0.2307692	0.422806	2	▇▁▁▁▁▁▁▂

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• is not a replication: 0
• is a replication: 1

replic_notes_DL

DL: coding notes/comments re new replication status variable

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

240 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	n_missing	complete_rate	n_unique	empty	min	max	whitespace
replic_notes_DL	DL: coding notes/comments re new replication status variable	character	240	0.0123457	3	0	68	115	0

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

is_replication_special_issue

whether paper is part of a special issue on replication studies

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

0 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
is_replication_special_issue	whether paper is part of a special issue on replication studies	haven_labelled	0. not part of replication special issue, 1. part of replication special issue	0	1	0	0	1	0.1069959	0.309746	2	▇▁▁▁▁▁▁▁

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• not part of replication special issue: 0
• part of replication special issue: 1

hypothesis_intro_contains_replic

whether hypothesis introduction phrase contains ‘replic’

Distribution

show_missing_values <- FALSE
if (has_labels(item)) {
  missing_values <- item[is.na(haven::zap_missing(item))]
  attributes(missing_values) <- attributes(item)
  if (!is.null(attributes(item)$labels)) {
    attributes(missing_values)$labels <- attributes(missing_values)$labels[is.na(attributes(missing_values)$labels)]
    attributes(item)$labels <- attributes(item)$labels[!is.na(attributes(item)$labels)]
  }
  if (is.double(item)) {
    show_missing_values <- length(unique(haven::na_tag(missing_values))) > 1
    item <- haven::zap_missing(item)
  }
  if (length(item_attributes$labels) == 0 && is.numeric(item)) {
    item <- haven::zap_labels(item)
  }
}
item_nomiss <- item[!is.na(item)]

# unnest mc_multiple and so on
if (
  is.character(item_nomiss) &&
  any(stringr::str_detect(item_nomiss, stringr::fixed(", "))) &&
  !is.null(item_info) &&
  (exists("type", item_info) && 
    any(stringr::str_detect(item_info$type, 
                            pattern = stringr::fixed("multiple"))))
  ) {
  item_nomiss <- unlist(stringr::str_split(item_nomiss, pattern = stringr::fixed(", ")))
}
attributes(item_nomiss) <- attributes(item)

old_height <- knitr::opts_chunk$get("fig.height")
non_missing_choices <- item_attributes[["labels"]]
many_labels <- length(non_missing_choices) > 7
go_vertical <- !is_numeric_or_time_var(item_nomiss) || many_labels
  
if ( go_vertical ) {
  # numeric items are plotted horizontally (because that's what usually expected)
  # categorical items are plotted vertically because we can use the screen real estate better this way

    if (is.null(choices) || 
        dplyr::n_distinct(item_nomiss) > length(non_missing_choices)) {
        non_missing_choices <- unique(item_nomiss)
        names(non_missing_choices) <- non_missing_choices
    }
  choice_multiplier <- old_height/6.5
    new_height <- 2 + choice_multiplier * length(non_missing_choices)
    new_height <- ifelse(new_height > 20, 20, new_height)
    new_height <- ifelse(new_height < 1, 1, new_height)
    if(could_disclose_unique_values(item_nomiss) && is.character(item_nomiss)) {
      new_height <- old_height
    }
    knitr::opts_chunk$set(fig.height = new_height)
}

wrap_at <- knitr::opts_chunk$get("fig.width") * 10

# todo: if there are free-text choices mingled in with the pre-defined ones, don't show
# todo: show rare items if they are pre-defined
# todo: bin rare responses into "other category"
if (!length(item_nomiss)) {
  cat("No non-missing values to show.")
} else if (!could_disclose_unique_values(item_nomiss)) {
  plot_labelled(item_nomiss, item_name, wrap_at, go_vertical)
} else {
  if (is.character(item_nomiss)) {
      char_count <- stringr::str_count(item_nomiss)
      attributes(char_count)$label <- item_label
      plot_labelled(char_count, 
                    item_name, wrap_at, FALSE, trans = "log1p", "characters")
  } else {
      cat(dplyr::n_distinct(item_nomiss), " unique, categorical values, so not shown.")
  }
}

knitr::opts_chunk$set(fig.height = old_height)

19 missing values.

Summary statistics

attributes(item) <- item_attributes
df = data.frame(item, stringsAsFactors = FALSE)
names(df) = html_item_name
escaped_table(codebook_table(df))

name	label	data_type	value_labels	n_missing	complete_rate	min	median	max	mean	sd	n_value_labels	hist
hypothesis_intro_contains_replic	whether hypothesis introduction phrase contains ‘replic’	haven_labelled	0. hypothesis intro does not contain ‘replic’, 1. hypothesis intro contains ‘replic’	19	0.9218107	0	0	1	0.0892857	0.2857944	2	▇▁▁▁▁▁▁▁

if (show_missing_values) {
  plot_labelled(missing_values, item_name, wrap_at)
}

if (!is.null(item_info)) {
  # don't show choices again, if they're basically same thing as value labels
  if (!is.null(choices) && !is.null(item_info$choices) && 
    all(names(na.omit(choices)) == item_info$choices) &&
    all(na.omit(choices) == names(item_info$choices))) {
    item_info$choices <- NULL
  }
  item_info$label_parsed <- 
    item_info$choice_list <- item_info$study_id <- item_info$id <- NULL
  pander::pander(item_info)
}

Value labels

if (!is.null(choices) && length(choices) && length(choices) < 30) {
    pander::pander(as.list(choices))
}

• hypothesis intro does not contain ‘replic’: 0
• hypothesis intro contains ‘replic’: 1

missingness_report

Missingness report

if (length(md_pattern)) {
  if (knitr::is_html_output()) {
    rmarkdown::paged_table(md_pattern, options = list(rows.print = 10))
  } else {
    knitr::kable(md_pattern)
  }
}

items

Codebook table

export_table(metadata_table)

jsonld

JSON-LD metadata

The following JSON-LD can be found by search engines, if you share this codebook publicly on the web.

{
  "name": "Positive Results in Standard vs Registered Reports",
  "description": "Coded variables to compare the proportion of papers that find support for their first tested hypothesis between Registered Reports and standard reports in Psychology\n\n\n## Table of variables\nThis table contains variable names, labels, and number of missing values.\nSee the complete codebook for more.\n\n[truncated]\n\n### Note\nThis dataset was automatically described using the [codebook R package](https://rubenarslan.github.io/codebook/) (version 0.8.2).",
  "creator": {
    "@type": "Person",
    "givenName": "Anne",
    "familyName": "Scheel",
    "affiliation": {
      "@type": "Organization",
      "name": "Eindhoven University of Technology"
    },
    "@type.1": "Person",
    "givenName.1": " Mitchell",
    "familiyName": "Schijen",
    "@type.2": "Person",
    "givenName.2": "Daniël",
    "familiyName.1": "Lakens"
  },
  "temporalCoverage": "Winter 2018",
  "datePublished": "2020-01-13",
  "keywords": ["id", "include_in_analysis", "exclusion_crit", "is_RR", "support", "support_binary", "is_replication", "contains_replic", "RR_hyp_intro_abstract", "RR_hyp_intro_fulltext", "hypothesis_quote", "result_quote", "hyp_from_fulltext", "result_from_fulltext", "doi", "title", "authors", "year", "journal", "abstract", "keywords", "coding_round", "combining_coding_notes", "uses_AS_coding", "coded_by_MS", "support_orig_MS", "support_revised_MS", "evaluation_certainty_MS", "hypothesis_MS", "hypothesis_introduction_MS", "finding_MS", "conclusion_MS", "hypothesis_from_fulltext_MS", "result_from_fulltext_MS", "fulltext_available_MS", "is_replication_old_MS", "is_original_old_MS", "time_spent_MS", "individual_coding_notes_MS", "coded_by_AS", "support_orig_AS", "support_revised_AS", "evaluation_certainty_AS", "hypothesis_AS", "hypothesis_introduction_AS", "finding_AS", "conclusion_AS", "hypothesis_from_fulltext_AS", "result_from_fulltext_AS", "fulltext_available_AS", "is_replication_old_AS", "is_original_old_AS", "time_spent_AS", "individual_coding_notes_AS", "is_replication_new_combining_notes", "is_replication_new_coded_by_AS", "is_replication_new_AS", "replic_notes_AS", "is_replication_new_coded_by_DL", "is_replication_new_DL", "replic_notes_DL", "is_replication_special_issue", "hypothesis_intro_contains_replic"],
  "@context": "http://schema.org/",
  "@type": "Dataset",
  "variableMeasured": [
    {
      "name": "id",
      "description": "ID number",
      "@type": "propertyValue"
    },
    {
      "name": "include_in_analysis",
      "description": "whether paper is included in final analysis",
      "value": "0. excluded,\n1. included",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "exclusion_crit",
      "description": "reason for exclusion",
      "@type": "propertyValue"
    },
    {
      "name": "is_RR",
      "description": "Registered Report (RR) or standard report (SR)",
      "value": "0. SR,\n1. RR",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "support",
      "description": "if the coded hypothesis received no support, partial support, or full support",
      "@type": "propertyValue"
    },
    {
      "name": "support_binary",
      "description": "whether the coded hypothesis was at least partially supported or not",
      "value": "0. no support,\n1. partial or full support",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_replication",
      "description": "whether the coded hypothesis is a direct replication of previous work",
      "value": "0. is not a replication,\n1. is a replication",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "contains_replic",
      "description": "whether the full text of the paper (excl. references) contains the string 'replic'",
      "value": "0. does not contain 'replic',\n1. contains 'replic'",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "RR_hyp_intro_abstract",
      "description": "phrase used to introduce/mark the coded hypothesis in RRs; from abstract",
      "@type": "propertyValue"
    },
    {
      "name": "RR_hyp_intro_fulltext",
      "description": "phrase used to introduce/mark the coded hypothesis in RRs; from full text",
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_quote",
      "description": "quote of the coded hypothesis",
      "@type": "propertyValue"
    },
    {
      "name": "result_quote",
      "description": "quote of the coded finding or conclusion",
      "@type": "propertyValue"
    },
    {
      "name": "hyp_from_fulltext",
      "description": "whether hypothesis was coded from abstract or full text",
      "value": "0. hypothesis from abstract alone,\n1. hypothesis from full text",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "result_from_fulltext",
      "description": "whether finding/conclusion was coded from abstract or full text",
      "value": "0. result from abstract alone,\n1. result from full text",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "doi",
      "description": "digital object identifier (DOI)",
      "@type": "propertyValue"
    },
    {
      "name": "title",
      "description": "title",
      "@type": "propertyValue"
    },
    {
      "name": "authors",
      "description": "author(s) of the paper",
      "@type": "propertyValue"
    },
    {
      "name": "year",
      "description": "publication year",
      "@type": "propertyValue"
    },
    {
      "name": "journal",
      "description": "journal the paper is published in",
      "@type": "propertyValue"
    },
    {
      "name": "abstract",
      "description": "abstract",
      "@type": "propertyValue"
    },
    {
      "name": "keywords",
      "description": "author keywords from meta-data",
      "@type": "propertyValue"
    },
    {
      "name": "coding_round",
      "description": "when the paper was coded (during piloting, in the main round, or as a replacement/late addition)\n",
      "value": "0. RR pilot,\n1. main coding round,\n2. SR replacements and late RR additions,\n3. second SR replacement,\n4. RR protocol replacement",
      "maxValue": 4,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "combining_coding_notes",
      "description": "Comments on how MS' and AS' codings were combined if there was disagreement",
      "@type": "propertyValue"
    },
    {
      "name": "uses_AS_coding",
      "description": "whether AS' or MS' coding was used for final analysis",
      "value": "0. uses MS coding,\n1. uses AS coding",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "coded_by_MS",
      "description": "whether MS coded the paper",
      "value": "0. not coded by MS,\n1. coded by MS",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "support_orig_MS",
      "description": "MS: whether hypothesis was supported (original verdict)",
      "@type": "propertyValue"
    },
    {
      "name": "support_revised_MS",
      "description": "MS: whether hypothesis was supported (revised verdict after discussion with MS)",
      "@type": "propertyValue"
    },
    {
      "name": "evaluation_certainty_MS",
      "description": "MS' coding certainty",
      "value": "1. not certain,\n2. ,\n3. ,\n4. ,\n5. very certain",
      "maxValue": 5,
      "minValue": 1,
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_MS",
      "description": "MS: hypothesis quote",
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_introduction_MS",
      "description": "MS: hypothesis introduction phrase",
      "@type": "propertyValue"
    },
    {
      "name": "finding_MS",
      "description": "MS: finding quote",
      "@type": "propertyValue"
    },
    {
      "name": "conclusion_MS",
      "description": "MS: conclusion quote",
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_from_fulltext_MS",
      "description": "MS: whether hypothesis was coded from abstract or full text",
      "value": "0. hypothesis from abstract alone,\n1. hypothesis from full text",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "result_from_fulltext_MS",
      "description": "MS: whether finding/conclusion was coded from abstract or full text",
      "value": "0. result from abstract alone,\n1. result from full text",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "fulltext_available_MS",
      "description": "MS: whether full text was available during coding",
      "value": "0. full text not available,\n1. full text available",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_old_MS",
      "description": "MS: whether paper contains a replication (old variable, not used!)",
      "value": "0. does not contain a replication,\n1. contains a replication",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_original_old_MS",
      "description": "MS: whether paper contains original research (old variable, not used!)",
      "value": "0. does not contain original work,\n1. contains original work",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "time_spent_MS",
      "description": "MS: time spent on coding the paper",
      "@type": "propertyValue"
    },
    {
      "name": "individual_coding_notes_MS",
      "description": "MS: coding notes/comments",
      "@type": "propertyValue"
    },
    {
      "name": "coded_by_AS",
      "description": "whether AS coded the paper",
      "value": "0. not coded by AS,\n1. coded by AS",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "support_orig_AS",
      "description": "AS: whether hypothesis was supported (original verdict)",
      "@type": "propertyValue"
    },
    {
      "name": "support_revised_AS",
      "description": "AS: whether hypothesis was supported (revised verdict after discussion with MS)",
      "@type": "propertyValue"
    },
    {
      "name": "evaluation_certainty_AS",
      "description": "AS' coding certainty",
      "value": "1. not certain,\n2. ,\n3. ,\n4. ,\n5. very certain",
      "maxValue": 5,
      "minValue": 1,
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_AS",
      "description": "AS: hypothesis quote",
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_introduction_AS",
      "description": "AS: hypothesis introduction phrase",
      "@type": "propertyValue"
    },
    {
      "name": "finding_AS",
      "description": "AS: finding quote",
      "@type": "propertyValue"
    },
    {
      "name": "conclusion_AS",
      "description": "AS: conclusion quote",
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_from_fulltext_AS",
      "description": "AS: whether hypothesis was coded from abstract or full text",
      "value": "0. hypothesis from abstract alone,\n1. hypothesis from full text",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "result_from_fulltext_AS",
      "description": "AS: whether finding/conclusion was coded from abstract or full text",
      "value": "0. result from abstract alone,\n1. result from full text",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "fulltext_available_AS",
      "description": "AS: whether full text was available during coding",
      "value": "0. full text not available,\n1. full text available",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_old_AS",
      "description": "AS: whether paper contains a replication (old variable, not used!)",
      "value": "0. does not contain a replication,\n1. contains a replication",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_original_old_AS",
      "description": "AS: whether paper contains original research (old variable, not used!)",
      "value": "0. does not contain original work,\n1. contains original work",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "time_spent_AS",
      "description": "AS: time spent on coding the paper",
      "@type": "propertyValue"
    },
    {
      "name": "individual_coding_notes_AS",
      "description": "AS: coding notes/comments",
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_new_combining_notes",
      "description": "Comments on how AS' and DL's new replication codings were combined if there was disagreement",
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_new_coded_by_AS",
      "description": "whether AS coded replication status (new variable) for the paper",
      "value": "0. replication status not coded by AS,\n1. replication status coded by AS",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_new_AS",
      "description": "AS: whether coded hypothesis is a direct replication of previous work",
      "value": "0. is not a replication,\n1. is a replication",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "replic_notes_AS",
      "description": "AS: coding notes/comments re new replication status variable",
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_new_coded_by_DL",
      "description": "whether DL coded replication status (new variable) for the paper",
      "value": "0. replication status not coded by DL,\n1. replication status coded by DL",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_new_DL",
      "description": "DL: whether coded hypothesis is a direct replication of previous work",
      "value": "0. is not a replication,\n1. is a replication",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "replic_notes_DL",
      "description": "DL: coding notes/comments re new replication status variable",
      "@type": "propertyValue"
    },
    {
      "name": "is_replication_special_issue",
      "description": "whether paper is part of a special issue on replication studies",
      "value": "0. not part of replication special issue,\n1. part of replication special issue",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    },
    {
      "name": "hypothesis_intro_contains_replic",
      "description": "whether hypothesis introduction phrase contains 'replic'",
      "value": "0. hypothesis intro does not contain 'replic',\n1. hypothesis intro contains 'replic'",
      "maxValue": 1,
      "minValue": 0,
      "@type": "propertyValue"
    }
  ]
}`